v1.1.0 – Extended simulation scenarios and reproducible analysis pipeline

Overview

This release updates the repository to reflect the analyses performed in the revised version of the manuscript comparing distance-based and model-based clustering methods for longitudinal psychiatric data.

The repository provides fully reproducible R scripts for both real data analyses and simulation experiments.

Major updates in this release

Simulation design extended

• Added simulation scenarios with unbalanced group-size distributions, reflecting heterogeneity commonly observed in psychiatric longitudinal cohorts.
• Implemented missing data mechanisms informed by the empirical missingness pattern observed in the PEPs cohort to increase the realism of the simulated datasets.

Improved computational pipeline

Simulation and model-fitting procedures were updated to support parallel execution using parLapply, substantially improving computational efficiency for large simulation experiments.

Repository reorganization

The repository structure has been reorganized to clearly separate real-data and simulation analyses:

1_Real_data_analysis
    1_1_functions.R
    1_2_analysis.R

2_Simulated_data_analysis
    2_1_functions.R
    2_2_analysis_balanced_clusters.R
    2_3_analysis_unbalanced_clusters.R

This structure improves transparency and reproducibility of the different analytical components.

Purpose of the repository

This repository supports methodological research on identifying trajectory-based response phenotypes in longitudinal psychiatric data and enables full reproducibility of the analyses reported in the accompanying manuscript.

Citation

If you use this code, please cite the Zenodo DOI associated with this release.